mastra-ai/mastra · error · MastraError

EXPERIMENT_NO_ITEMS

EXPERIMENT_NO_ITEMS

Error message

Cannot run experiment: dataset "${this.id}" has no items at version ${targetVersion}

What it means

Thrown by Dataset.startExperimentAsync when the datasets store returns zero items for the dataset at the resolved target version. An experiment with no items to run would produce empty results, so Mastra refuses to start it. The target version defaults to the dataset's current version unless config.version is given.

Source

Thrown at packages/core/src/datasets/dataset.ts:395

    const dataset = await datasetsStore.getDatasetById({ id: this.id, filters: this.#scope });
    if (!dataset) {
      throw new MastraError({
        id: 'DATASET_NOT_FOUND',
        text: `Dataset not found: ${this.id}`,
        domain: 'STORAGE',
        category: 'USER',
      });
    }

    // Validate that dataset has items before creating experiment record
    const targetVersion = config.version ?? dataset.version;
    const items = await datasetsStore.getItemsByVersion({
      datasetId: this.id,
      version: targetVersion,
    });
    if (items.length === 0) {
      throw new MastraError({
        id: 'EXPERIMENT_NO_ITEMS',
        text: `Cannot run experiment: dataset "${this.id}" has no items at version ${targetVersion}`,
        domain: 'STORAGE',
        category: 'USER',
      });
    }

    const experimentId = crypto.randomUUID();

    if (experimentsStore) {
      await experimentsStore.createExperiment({
        id: experimentId,
        datasetId: this.id,
        datasetVersion: targetVersion,
        targetType: config.targetType ?? 'agent',
        targetId: config.targetId ?? 'inline',
        totalItems: items.length,
        name: config.name,

View on GitHub (pinned to 75dd419e61)

Solutions

1. Check the dataset actually has items at the target version (listItems / getItemsByVersion) before starting
2. Omit config.version to use the dataset's current version, or pass the correct version that contains items
3. Add items to the dataset at the desired version before creating the experiment

Example fix

// before
await dataset.startExperimentAsync({ version: 3, targetId: 'myAgent' });
// after
const items = await dataset.getItemsByVersion({ datasetId: dataset.id, version: 3 });
if (items.length > 0) await dataset.startExperimentAsync({ version: 3, targetId: 'myAgent' });

const items = await datasetsStore.getItemsByVersion({ datasetId: dataset.id, version: config.version ?? dataset.version });
if (!items.length) throw new Error(`Dataset ${dataset.id} has no items at version ${config.version ?? dataset.version}`);

function hasItems<T>(items: T[] | undefined | null): items is T[] { return Array.isArray(items) && items.length > 0; }

try {
  await dataset.startExperimentAsync(config);
} catch (e) {
  if (isMastraError(e) && e.id === 'EXPERIMENT_NO_ITEMS') {
    console.warn(`Skipping experiment: no items at version ${config.version ?? 'current'}`);
  } else throw e;
}

When it happens

Trigger: Calling startExperimentAsync on a dataset whose items were never added, or at a config.version that predates any items (or after items were all removed from that version).

Common situations: Typo in version number; dataset created but items not yet inserted; items deleted from an old version; pointing at version 1 of a dataset whose data only exists in version 2; empty dataset in a fresh environment.

Related errors

• EXPERIMENT_NO_ITEMS
• EXPERIMENT_NOT_FOUND
• EXPERIMENT_RESULT_MISSING_EXPERIMENT_ID
• EXPERIMENT_INVALID_TARGET
• EXPERIMENT_TARGET_NOT_FOUND

AI-assisted analysis of mastra-ai/mastra@75dd419e61 (2026-08-30). Data as JSON: /api/errors/b9dfa697275d1ad6. Report an issue: GitHub.